Abstract

Community structures are integral and independent parts in a network. Community detection plays an important role in social networks for understanding the structure and predicting user behaviors. Many algorithms have been devised for accurate and efficient community detecting, but there are few community detection algorithms using node similarity. In most real-world networks, nodes tend to create tightly knit groups characterized by a relatively high density of ties. The higher the clustering coefficient of a node, the more aggregative the neighboring nodes are. In this paper, we propose an adjacent node similarity optimization combination connectivity algorithm (ASOCCA) for accurate community detection. ASOCCA utilizes the local similarity measure based on clustering coefficient to identify the closest neighbors of each node, then obtains several sets of connected components by combining different pairs of nodes, and finally forms initial communities. In addition, the community merging strategy is applied to further optimize the community structure. To evaluate the performance of the proposed algorithm, six real-world networks and two LFR networks with diverse network size are used to compare ASOCCA with five state-of-the-art community detection algorithms. The experimental results show that ASOCCA achieves better detection accuracy than several existing algorithms.

Highlights

  • The popularity of social networks such as Facebook, Twitter, and Weibo have produced tremendous network data and how to analyze these complex-structured data and derive useful information becomes a hot research topic [1]

  • In this paper, based on the similarity of adjacent nodes, we propose a novel community detection algorithm named Adjacent node Similarity Optimization Combination Connectivity Algorithm, which focuses on the non-overlapping community detection for undirected and unweighted network

  • The superior performance is attributed to the local information based local similarity and the merging strategy proposed in this paper

Read more

Summary

INTRODUCTION

The popularity of social networks such as Facebook, Twitter, and Weibo have produced tremendous network data and how to analyze these complex-structured data and derive useful information becomes a hot research topic [1]. From the perspective of structural similarity, the study of community division has received increasing attention from scholars, and many effective and powerful approaches have been developed [13]–[23]. These algorithms firstly compute similarity, either between the nodes or edges of a network, and utilize a variety of aggregation strategies to detect communities. Walktrap algorithm [17], [18] derives similarity based on random-walks between nodes and performs hierarchical aggregation to detect communities. Label Propagation Algorithm (LPA) [19] uses local similarity, which is based on the concept of information diffusion in networks.

RELATED WORK
MODULARITY Q
1: Initialization
CONCLUSION
Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call